What this reads, and what it never does.

CareerOps is operated by Mike Miller as an independent product. There is no company behind it, no advertising in it, and no third party that receives your data for their own purposes. Questions go to mike@mikemiller.ai.

Only what a job search is made of, and only what you connect:

• Your account. Email address, display name, and a hashed password or your Google sign-in identifier.
• Your job search. Opportunities, companies, people, applications, interviews, notes, commitments and the timeline that connects them.
• Job-search mail. Messages matched by a narrow Gmail query, stored so they can be attached to the right process and so a reply can be drafted for them.
• Documents you upload. Resumes and cover letters, kept version by version so an application stays attached to the exact file that was sent.
• Interview recordings you already have. Notes and transcripts read from Granola, when you connect it.

• Your calendar. A sync reads events, decides locally which are recruiting conversations, and keeps only the interviews. Every other event is counted and dropped — nothing from it is written down.
• Your OAuth tokens. They live in AWS Secrets Manager, and the database holds only a reference to them. A database backup does not contain the credential that reads your mail.
• Payment details. There is nothing to pay for and no payment processor involved.
• Tracking. No analytics scripts, no advertising pixels, no third-party cookies. The only cookies are your session and your time zone.

When you connect Gmail, Calendar or Granola, CareerOps requests the narrowest scopes that make the feature work.

• Gmail — read and compose. Reading is what lets a message be classified and attached to an opportunity. Compose is what lets a draft be saved for you to review. Google publishes no draft-only scope, so the guarantee that nothing is sent does not come from OAuth — it comes from there being no send path in the code, and from an approval gate that refuses the action at every confidence level.
• Calendar — read only. CareerOps cannot create, move or delete an event.

CareerOps’ use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements. Google user data is used only to provide the features described here. It is never sold, never used for advertising, and never used to train any model. No human reads it except you.

Disconnecting an account from Settings revokes the grant and deletes the stored credential immediately.

Model calls go to Anthropic to classify a message, summarise an interview, estimate role fit, or draft a reply. Only the text needed for that specific task is sent. Anthropic does not train on data submitted through its API.

Every call is recorded in your own workspace with its purpose, its cost and its outcome — you can see the whole bill and every agent that spent it under Settings. A model never moves a stage, never computes a priority, and never sends anything; it proposes, and you decide.

On Amazon Web Services in the US East (N. Virginia) region, in an Aurora PostgreSQL database that is encrypted at rest and reachable only over TLS. Every table that belongs to a workspace is protected by PostgreSQL row-level security, so isolation is enforced by the database rather than by application code remembering to ask.

Data is kept while your workspace exists. Anything in it can be deleted from the interface at any time — an opportunity, a document, a message, a raw interview note. Email mike@mikemiller.ai to have the entire workspace deleted, and it will be removed within seven days along with every credential attached to it.

If this policy changes in a way that affects what is collected or how it is used, the date at the top changes and you will be told before it takes effect. See also the terms and the security notes.